AI Tools Preset Containers

AI Tools Preset Containers provide data scientists and developers with ready-to-use environments for tasks like data analysis, data processing, machine learning, and deep learning. Each container includes Python* packages and tools tailored to specific tasks. More details are provided in the table below.

Preset Containers

Preset Container Name	Purpose	Tools	Image Name
Classical Machine Learning	Accelerate your machine learning and data science pipelines with the power of open libraries optimized for Intel® architectures	Intel® Extension for Scikit-learn*, Intel® Optimization for XGBoost*, Modin*	intel/classical-ml:latest-py3.11
Deep Learning PyTorch* CPU	Boost the performance of your workloads, reduce model size, and improve the speed of your Deep Learning deployments on Intel® Xeon® processors with Intel® Extension for PyTorch*	Intel® Extension for PyTorch, Intel® Neural Compressor, ONNX Runtime*	intel/deep-learning:pytorch-latest-py3.11
Deep Learning PyTorch* GPU	Boost the performance of your workloads, reduce model size, and improve the speed of your Deep Learning deployments on Intel® Data Center GPU Max Series with Intel® Extension for PyTorch*	Intel® Extension for PyTorch, Intel® Neural Compressor, DeepSpeed*	intel/deep-learning:pytorch-gpu-latest-py3.11
Deep Learning TensorFlow* CPU	Boost the performance of your workloads, reduce model size, and improve the speed of your Deep Learning deployments on Intel® Xeon® processors with Intel® Extension for TensorFlow*	Intel® Extension for Tensorflow, Intel® Neural Compressor, ONNX Runtime*	intel/deep-learning:tensorflow-latest-py3.11
Deep Learning TensorFlow* GPU	Boost the performance of your workloads, reduce model size, and improve the speed of your Deep Learning deployments on Intel® Data Center GPU Max Series with Intel® Extension for TensorFlow*	Intel® Extension for Tensorflow, Intel® Optimization for Horovod*, Intel® Neural Compressor	intel/deep-learning:tensorflow-gpu-latest-py3.11
Deep Learning JAX* CPU	Reduce model size and improve the speed of your Deep Learning deployments on Intel® Xeon® processors with JAX*	JAX*, Intel® Distribution of Python	intel/deep-learning:jax-latest-py3.11

Prerequisites

1. Make sure Docker is installed on the machine. Follow the instruction here to install docker engine on a host machine.

2. Pull a preset container of your choice from the AI Tools Selector or from the table. The commands below use the Deep Learning for PyTorch* CPU preset as an example.

docker pull intel/deep-learning:pytorch-latest-py311

Run Preset Container

There are three modes to run these containers:

• Jupyter Notebook
• Interactive
• Multi-Node Distributed Training (Deep Learning PyTorch GPU Deep Learning TensorFlow GPU)

Note

Modify the commands below to fit your use case, including the image, environment variables, and GPU device path.

Run using Jupyter Notebook

This mode starts a JupyterLab notebook server, accessible via a web browser. Each container includes a Jupyter kernel to enable the conda environment in the Jupyter notebook. The port for this server is 8888 and is exposed by default when you run the container.

Note

When launching a Jupyter notebook server this way, Docker assigns a network that allows the container to communicate with other applications, such as a web browser. By default, Docker uses the bridge network. If you are trying to access this server from a remote machine via SSH, change the network mode with the --net=host flag and enable local port forwarding with ssh -L 8888:8888.

Run on Jupyter CPU

docker run -it --rm \
    -p 8888:8888 --shm-size=12G \
    -v ${PWD}:/home/dev/workdir \
    intel/deep-learning:pytorch-latest-py3.11

Note

Some applications use shared memory to share data between processes. The default shared memory segment size for Docker containers is 64MB, which may be insufficient for multithreaded applications (for example, Modin*). Increase the shared memory size using --shm-size.

Run on Jupyter GPU

Identify your system's RENDER and VIDEO group values to enable Intel® Data Center GPU Flex/Max Series:

RENDER=$(getent group render | sed -E 's,^render:[^:]*:([^:]*):.*$,\1,')
VIDEO=$(getent group video | sed -E 's,^video:[^:]*:([^:]*):.*$,\1,')
test -z "$RENDER" || RENDER_GROUP="--group-add ${RENDER}"
test -z "$VIDEO" || VIDEO_GROUP="--group-add ${VIDEO}"

Run the container:

docker run -it --rm \
    ${RENDER_GROUP} \
    ${VIDEO_GROUP} \
    --device=/dev/dri \
    -p 8888:8888  --shm-size=12G \
    -v ${PWD}:/home/dev/workdir \
    -v /dev/dri/by-path:/dev/dri/by-path \
    intel/deep-learning:pytorch-gpu-latest-py3.11

Next Steps

1. After running the command, the terminal should display output similar to the image below . Use the displayed server address and port (for example, http://127.0.0.1:8888) to connect to the Jupyter server in a web browser. The token displayed after token= can be used as a password to log in. For example, in the image displayed above the token is b66e74a85bc2570bf15782e5124c933c3a4ddabd2cf2d7d3.

2. Select a notebook sample from the Overview notebook found in directory where you launched the server. In this example, the intel/deep-learning:pytorch-latest-py3.11 container has a notebook titled Deep_Learning_Samples_Overview.ipynb when launched in Jupyter mode.

3. After selecting a notebook sample, choose the appropriate kernel from the dropdown menu available when loading the notebook. For each container, there is a kernel with the environment name: classical-ml, jax-cpu, pytorch-cpu, pytorch-gpu, tensorflow-cpu, and tensorflow-gpu.

Advanced Jupyter Server Configuration

Modify your notebook server command by using the default example below to change the network (port/ip) and security (privilege) settings by appending it to the Docker run commands above:

docker run ... intel/deep-learning:pytorch-latest-py3.11 \
    bash -c "jupyter notebook --notebook-dir=~/jupyter \
            --port 8888 \
            --ip 0.0.0.0 \
            --no-browser \
            --allow-root"

Run in Interactive Mode

This mode provides an interactive shell for working directly with the container's bash environment. Use this command to start the container in an interactive mode:

Run on CPU

docker run -it --rm \
    -p 8888:8888 --shm-size=12G \
    -v ${PWD}:/home/dev/workdir \
    intel/deep-learning:pytorch-latest-py3.11 bash

Run on GPU

Identify your system's RENDER and VIDEO group values to enable Intel® Data Center GPU Flex/Max Series:

RENDER=$(getent group render | sed -E 's,^render:[^:]*:([^:]*):.*$,\1,')
VIDEO=$(getent group video | sed -E 's,^video:[^:]*:([^:]*):.*$,\1,')
test -z "$RENDER" || RENDER_GROUP="--group-add ${RENDER}"
test -z "$VIDEO" || VIDEO_GROUP="--group-add ${VIDEO}"

docker run -it --rm \
    ${RENDER_GROUP} \
    ${VIDEO_GROUP} \
    --device=/dev/dri \
    -p 8888:8888 --shm-size=12G \
    -v ${PWD}:/home/dev/workdir \
    -v /dev/dri/by-path:/dev/dri/by-path \
    intel/deep-learning:pytorch-latest-py3.11 bash

Note

Some applications use shared memory to share data between processes. The default shared memory segment size for Docker containers is 64MB, which may be insufficient for multithreaded applications (for example, Modin*). Increase the shared memory size using --shm-size.

Next Steps

1. For each container, there is a kernel with the environment name: classical-ml, jax, pytorch-cpu, pytorch-gpu, tensorflow-cpu, and tensorflow-gpu. Use the command below to activate the pytorch-cpu environment in the Deep Learning PyTorch* CPU preset:

   conda activate pytorch-cpu
   

2. Select a test from the sample-tests folder and run it using the following command as an example:

   python sample-tests/intel_extension_for_pytorch/test_ipex.py
   

   The test_ipex.py script utilizes PyTorch* to classify images with a pre-trained ResNet-50 model.
   Users can specify whether to run the script on a CPU or XPU, and there is an option to apply optimizations using Intel® Extension for PyTorch*.
   The script transfers both the model and data to the chosen device, and then measures the average inference time over 100 runs, excluding the initial warm-up phase.
   Finally, the script prints the average inference time.

Note

The sample-tests folder may vary by container. Some tests use a bash script.

Run in Multi-Node Distributed Mode [Advanced]

Follow the instructions provided for Intel® Extension for TensorFlow* or Intel® Extension for PyTorch* using the Deep Learning TensorFlow* GPU or Deep Learning PyTorch* GPU presets, respectively.

Troubleshooting and Support

If you face any issues while using the container, refer to the troubleshooting section. For additional assistance, you can submit an issue.

---

*Other names and brands may be claimed as the property of others. Trademarks